To achieve true continuous delivery, teams need to leverage end-to-end automation across their deployment pipelines and tooling. Deployment health checks today involve someone pinging the production app and saying "Yep, the app is still up--everything is OK!" In short, availability = success.
In reality, availability doesn't equal success, and the only way to get a true measure of health is to have multiple engineers sniff-test production by grep'ing logs or looking at monitoring charts. It's not unusual for teams to have four or five engineers spending 60 to 120 minutes per deployment making sure everything really is OK. But what if AI/ML could automate this process?
This session will take a look at some of the machine learning techniques you can apply to automate deployment verification and health checks.